Governing Artificial Intelligence in Europe: Policy Frames, Implementation Gaps, and the Task–Skill–Institution Model

This study analyses how EU member states formulate and implement national artificial intelligence (AI) policies, investigating the relationships between policy frames, implementation practices and innovation and labor market outcomes. Based on a corpus of national AI strategies of EU member states, we apply Structural Topic Modeling combined with frame analysis to identify dominant discursive orientations and their temporal evolution. The analysis reveals a reconfiguration of priorities: the emphasis on capitalizing on economic opportunities is decreasing, while concerns about governance, compliance and risk management are gaining weight, and the capacity development framework remains relatively stable. We further investigate the extent to which the prevalence of these frames is associated with implementation tools and indicators of innovation performance, including R&D dynamics. In addition, we introduce the Task–Skill–Institution (TSI) model, an integrated conceptual framework that conceptualizes AI-induced labor market vulnerabilities as emergent outcomes of interactions between exposure to automation, skill elasticity, and institutional capacities. The results indicate that cross-national variations in the maturity of AI ecosystems can be partly explained by the gap between strategic discourse and implementation capacity, as well as by systemic configurations that mediate technological impact. The study provides an empirical and conceptual basis for assessing the alignment between discourse, governance, and implementation, and formulates directions for strengthening AI policies in Europe.